Rollup merge of rust-lang#82415 - petrochenkov:modin3, r=davidtwco

expand: Refactor module loading

This is an accompanying PR to rust-lang#82399, but they can be landed independently.
See individual commits for more details.

Anyone should be able to review this equally well because all people actually familiar with this code left the project.

Author: Dylan-DPC

compiler/rustc_builtin_macros/src/source_util.rs

@@ -4,7 +4,7 @@ use rustc_ast::token;
 use rustc_ast::tokenstream::TokenStream;
 use rustc_ast_pretty::pprust;
 use rustc_expand::base::{self, *};
-use rustc_expand::module::DirectoryOwnership;
+use rustc_expand::module::DirOwnership;
 use rustc_parse::parser::{ForceCollect, Parser};
 use rustc_parse::{self, new_parser_from_file};
 use rustc_session::lint::builtin::INCOMPLETE_INCLUDE;
@@ -101,7 +101,7 @@ pub fn expand_include<'cx>(
         None => return DummyResult::any(sp),
     };
     // The file will be added to the code map by the parser
-    let mut file = match cx.resolve_path(file, sp) {
+    let file = match cx.resolve_path(file, sp) {
         Ok(f) => f,
         Err(mut err) => {
             err.emit();
@@ -114,10 +114,9 @@ pub fn expand_include<'cx>(
     // then the path of `bar.rs` should be relative to the directory of `file`.
     // See https://github.com/rust-lang/rust/pull/69838/files#r395217057 for a discussion.
     // `MacroExpander::fully_expand_fragment` later restores, so "stack discipline" is maintained.
-    file.pop();
-    cx.current_expansion.directory_ownership = DirectoryOwnership::Owned { relative: None };
-    let mod_path = cx.current_expansion.module.mod_path.clone();
-    cx.current_expansion.module = Rc::new(ModuleData { mod_path, directory: file });
+    let dir_path = file.parent().unwrap_or(&file).to_owned();
+    cx.current_expansion.module = Rc::new(cx.current_expansion.module.with_dir_path(dir_path));
+    cx.current_expansion.dir_ownership = DirOwnership::Owned { relative: None };
 
     struct ExpandResult<'a> {
         p: Parser<'a>,

compiler/rustc_expand/src/base.rs

@@ -1,11 +1,11 @@
 use crate::expand::{self, AstFragment, Invocation};
-use crate::module::DirectoryOwnership;
+use crate::module::DirOwnership;
 
 use rustc_ast::ptr::P;
 use rustc_ast::token::{self, Nonterminal};
 use rustc_ast::tokenstream::{CanSynthesizeMissingTokens, LazyTokenStream, TokenStream};
 use rustc_ast::visit::{AssocCtxt, Visitor};
-use rustc_ast::{self as ast, AstLike, Attribute, NodeId, PatKind};
+use rustc_ast::{self as ast, AstLike, Attribute, Item, NodeId, PatKind};
 use rustc_attr::{self as attr, Deprecation, Stability};
 use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::sync::{self, Lrc};
@@ -900,21 +900,40 @@ pub trait ResolverExpand {
     fn cfg_accessible(&mut self, expn_id: ExpnId, path: &ast::Path) -> Result<bool, Indeterminate>;
 }
 
-#[derive(Clone)]
+#[derive(Clone, Default)]
 pub struct ModuleData {
+    /// Path to the module starting from the crate name, like `my_crate::foo::bar`.
     pub mod_path: Vec<Ident>,
-    pub directory: PathBuf,
+    /// Stack of paths to files loaded by out-of-line module items,
+    /// used to detect and report recursive module inclusions.
+    pub file_path_stack: Vec<PathBuf>,
+    /// Directory to search child module files in,
+    /// often (but not necessarily) the parent of the top file path on the `file_path_stack`.
+    pub dir_path: PathBuf,
+}
+
+impl ModuleData {
+    pub fn with_dir_path(&self, dir_path: PathBuf) -> ModuleData {
+        ModuleData {
+            mod_path: self.mod_path.clone(),
+            file_path_stack: self.file_path_stack.clone(),
+            dir_path,
+        }
+    }
 }
 
 #[derive(Clone)]
 pub struct ExpansionData {
     pub id: ExpnId,
     pub depth: usize,
     pub module: Rc<ModuleData>,
-    pub directory_ownership: DirectoryOwnership,
+    pub dir_ownership: DirOwnership,
     pub prior_type_ascription: Option<(Span, bool)>,
 }
 
+type OnExternModLoaded<'a> =
+    Option<&'a dyn Fn(Ident, Vec<Attribute>, Vec<P<Item>>, Span) -> (Vec<Attribute>, Vec<P<Item>>)>;
+
 /// One of these is made during expansion and incrementally updated as we go;
 /// when a macro expansion occurs, the resulting nodes have the `backtrace()
 /// -> expn_data` of their expansion context stored into their span.
@@ -932,15 +951,15 @@ pub struct ExtCtxt<'a> {
     /// Called directly after having parsed an external `mod foo;` in expansion.
     ///
     /// `Ident` is the module name.
-    pub(super) extern_mod_loaded: Option<&'a dyn Fn(&ast::Crate, Ident)>,
+    pub(super) extern_mod_loaded: OnExternModLoaded<'a>,
 }
 
 impl<'a> ExtCtxt<'a> {
     pub fn new(
         sess: &'a Session,
         ecfg: expand::ExpansionConfig<'a>,
         resolver: &'a mut dyn ResolverExpand,
-        extern_mod_loaded: Option<&'a dyn Fn(&ast::Crate, Ident)>,
+        extern_mod_loaded: OnExternModLoaded<'a>,
     ) -> ExtCtxt<'a> {
         ExtCtxt {
             sess,
@@ -952,8 +971,8 @@ impl<'a> ExtCtxt<'a> {
             current_expansion: ExpansionData {
                 id: ExpnId::root(),
                 depth: 0,
-                module: Rc::new(ModuleData { mod_path: Vec::new(), directory: PathBuf::new() }),
-                directory_ownership: DirectoryOwnership::Owned { relative: None },
+                module: Default::default(),
+                dir_ownership: DirOwnership::Owned { relative: None },
                 prior_type_ascription: None,
             },
             force_mode: false,

compiler/rustc_expand/src/expand.rs

@@ -3,7 +3,7 @@ use crate::config::StripUnconfigured;
 use crate::configure;
 use crate::hygiene::SyntaxContext;
 use crate::mbe::macro_rules::annotate_err_with_kind;
-use crate::module::{parse_external_mod, push_directory, Directory, DirectoryOwnership};
+use crate::module::{mod_dir_path, parse_external_mod, DirOwnership, ParsedExternalMod};
 use crate::placeholders::{placeholder, PlaceholderExpander};
 
 use rustc_ast as ast;
@@ -355,16 +355,17 @@ impl<'a, 'b> MacroExpander<'a, 'b> {
     // FIXME: Avoid visiting the crate as a `Mod` item,
     // make crate a first class expansion target instead.
     pub fn expand_crate(&mut self, mut krate: ast::Crate) -> ast::Crate {
-        let mut module = ModuleData {
-            mod_path: vec![Ident::from_str(&self.cx.ecfg.crate_name)],
-            directory: match self.cx.source_map().span_to_unmapped_path(krate.span) {
-                FileName::Real(name) => name.into_local_path(),
-                other => PathBuf::from(other.to_string()),
-            },
+        let file_path = match self.cx.source_map().span_to_unmapped_path(krate.span) {
+            FileName::Real(name) => name.into_local_path(),
+            other => PathBuf::from(other.to_string()),
         };
-        module.directory.pop();
-        self.cx.root_path = module.directory.clone();
-        self.cx.current_expansion.module = Rc::new(module);
+        let dir_path = file_path.parent().unwrap_or(&file_path).to_owned();
+        self.cx.root_path = dir_path.clone();
+        self.cx.current_expansion.module = Rc::new(ModuleData {
+            mod_path: vec![Ident::from_str(&self.cx.ecfg.crate_name)],
+            file_path_stack: vec![file_path],
+            dir_path,
+        });
 
         let krate_item = AstFragment::Items(smallvec![P(ast::Item {
             attrs: krate.attrs,
@@ -1245,10 +1246,12 @@ impl<'a, 'b> MutVisitor for InvocationCollector<'a, 'b> {
     }
 
     fn visit_block(&mut self, block: &mut P<Block>) {
-        let old_directory_ownership = self.cx.current_expansion.directory_ownership;
-        self.cx.current_expansion.directory_ownership = DirectoryOwnership::UnownedViaBlock;
+        let orig_dir_ownership = mem::replace(
+            &mut self.cx.current_expansion.dir_ownership,
+            DirOwnership::UnownedViaBlock,
+        );
         noop_visit_block(block, self);
-        self.cx.current_expansion.directory_ownership = old_directory_ownership;
+        self.cx.current_expansion.dir_ownership = orig_dir_ownership;
     }
 
     fn flat_map_item(&mut self, item: P<ast::Item>) -> SmallVec<[P<ast::Item>; 1]> {
@@ -1276,63 +1279,76 @@ impl<'a, 'b> MutVisitor for InvocationCollector<'a, 'b> {
                 })
             }
             ast::ItemKind::Mod(_, ref mut mod_kind) if ident != Ident::invalid() => {
-                let sess = &self.cx.sess.parse_sess;
-                let orig_ownership = self.cx.current_expansion.directory_ownership;
-                let mut module = (*self.cx.current_expansion.module).clone();
-
-                let pushed = &mut false; // Record `parse_external_mod` pushing so we can pop.
-                let dir = Directory { ownership: orig_ownership, path: module.directory };
-                let Directory { ownership, path } = match mod_kind {
-                    ModKind::Loaded(_, Inline::Yes, _) => {
+                let (file_path, dir_path, dir_ownership) = match mod_kind {
+                    ModKind::Loaded(_, inline, _) => {
                         // Inline `mod foo { ... }`, but we still need to push directories.
+                        assert!(
+                            *inline == Inline::Yes,
+                            "`mod` item is loaded from a file for the second time"
+                        );
+                        let (dir_path, dir_ownership) = mod_dir_path(
+                            &self.cx.sess,
+                            ident,
+                            &attrs,
+                            &self.cx.current_expansion.module,
+                            self.cx.current_expansion.dir_ownership,
+                        );
                         item.attrs = attrs;
-                        push_directory(&self.cx.sess, ident, &item.attrs, dir)
-                    }
-                    ModKind::Loaded(_, Inline::No, _) => {
-                        panic!("`mod` item is loaded from a file for the second time")
+                        (None, dir_path, dir_ownership)
                     }
                     ModKind::Unloaded => {
                         // We have an outline `mod foo;` so we need to parse the file.
-                        let (items, inner_span, dir) =
-                            parse_external_mod(&self.cx.sess, ident, span, dir, &mut attrs, pushed);
+                        let old_attrs_len = attrs.len();
+                        let ParsedExternalMod {
+                            mut items,
+                            inner_span,
+                            file_path,
+                            dir_path,
+                            dir_ownership,
+                        } = parse_external_mod(
+                            &self.cx.sess,
+                            ident,
+                            span,
+                            &self.cx.current_expansion.module,
+                            self.cx.current_expansion.dir_ownership,
+                            &mut attrs,
+                        );
 
-                        let krate =
-                            ast::Crate { attrs, items, span: inner_span, proc_macros: vec![] };
                         if let Some(extern_mod_loaded) = self.cx.extern_mod_loaded {
-                            extern_mod_loaded(&krate, ident);
+                            (attrs, items) = extern_mod_loaded(ident, attrs, items, inner_span);
                         }
 
-                        *mod_kind = ModKind::Loaded(krate.items, Inline::No, inner_span);
-                        item.attrs = krate.attrs;
-                        // File can have inline attributes, e.g., `#![cfg(...)]` & co. => Reconfigure.
-                        item = match self.configure(item) {
-                            Some(node) => node,
-                            None => {
-                                if *pushed {
-                                    sess.included_mod_stack.borrow_mut().pop();
-                                }
-                                return Default::default();
-                            }
-                        };
-                        dir
+                        *mod_kind = ModKind::Loaded(items, Inline::No, inner_span);
+                        item.attrs = attrs;
+                        if item.attrs.len() > old_attrs_len {
+                            // If we loaded an out-of-line module and added some inner attributes,
+                            // then we need to re-configure it.
+                            // FIXME: Attributes also need to be recollected
+                            // for resolution and expansion.
+                            item = configure!(self, item);
+                        }
+                        (Some(file_path), dir_path, dir_ownership)
                     }
                 };
 
                 // Set the module info before we flat map.
-                self.cx.current_expansion.directory_ownership = ownership;
-                module.directory = path;
+                let mut module = self.cx.current_expansion.module.with_dir_path(dir_path);
                 module.mod_path.push(ident);
+                if let Some(file_path) = file_path {
+                    module.file_path_stack.push(file_path);
+                }
+
                 let orig_module =
                     mem::replace(&mut self.cx.current_expansion.module, Rc::new(module));
+                let orig_dir_ownership =
+                    mem::replace(&mut self.cx.current_expansion.dir_ownership, dir_ownership);
 
                 let result = noop_flat_map_item(item, self);
 
                 // Restore the module info.
+                self.cx.current_expansion.dir_ownership = orig_dir_ownership;
                 self.cx.current_expansion.module = orig_module;
-                self.cx.current_expansion.directory_ownership = orig_ownership;
-                if *pushed {
-                    sess.included_mod_stack.borrow_mut().pop();
-                }
+
                 result
             }
             _ => {

compiler/rustc_expand/src/lib.rs

@@ -1,5 +1,7 @@
+#![feature(bool_to_option)]
 #![feature(crate_visibility_modifier)]
 #![feature(decl_macro)]
+#![feature(destructuring_assignment)]
 #![feature(or_patterns)]
 #![feature(proc_macro_diagnostic)]
 #![feature(proc_macro_internals)]